I understand that encoding/decoding can be optional but I think that in
some cases they can help to simplify the code.
For example, in the last days I developed a tcp server (using
ProtocolHandler) that read ReadMessage objects and write WriteMessage
objects.
Hence I wrote a TcpProtocolEncoder that trasform WriteMessage to
ByteBuffer and a TcpProtocolDecoder that trasform ByteBuffer to ReadMessage.
Now I would like to add the vm pipe transport type support to the same
server but because I can't use encoding/decoding, MINA call
ProtocolHandler.messageReceived() with a message of type WriteMessage.
So when receiving messages from a vm pipe client the messageReceived()
should first trasform the WriteMessage in a ReadMessage.
Hence the messageReceived() code depends on the session trasport type.
My question is: is there a design flaw in my project or I missed
something in the use of vm pipe trasport type?
Paolo
Jose Alberto Fernandez ha scritto:
There is no serialization necessary when doing things in memory; hence
encoding/decoding is not needed.
Jose
-----Original Message-----
From: Paolo Perrucci [mailto:[EMAIL PROTECTED]
Sent: 17 October 2005 16:49
To: [email protected]
Subject: Protocol decoder for vm pipe session
Hi all,
it seems that vm pipe based protocol session doesn't use protocol
endoder and decoder. Indeed ProtocolSession.getEncoder() return null.
Is this a bug or a feature ?
Thanks
Paolo
